Change GetVolume() to be asynchronous when being called by pulse thread

This patch changes GetVolume() to be asynchronous when being called by pulse thread.

Pulseaudio does not allow calling pa_threaded_mainloop_lock() and pa_threaded_mainloop_wait() in its callback thread.
So the GetVolume() has to be asynchronous when being called in the pulse thread, and synchronous when being called by other threads.


BUG=178057
TEST=enable pulse, and make a call using https://webrtc-demos.appspot.com/html/pc1.html

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=184629

[no longer working on chromium, 2013-02-25 12:36:32]

Dale, could you please review this CL?

[DaleCurtis, 2013-02-25 18:52:18]

lgtm % comment nits.

https://codereview.chromium.org/12310102/diff/1/media/audio/pulse/pulse_input.cc
File media/audio/pulse/pulse_input.cc (right):

https://codereview.chromium.org/12310102/diff/1/media/audio/pulse/pulse_input...
media/audio/pulse/pulse_input.cc:281: stream->volume_ =
static_cast<double>(volume);
I'd add a comment here that this is thread safe because SetVolume runs under a
PulseLock.

https://codereview.chromium.org/12310102/diff/1/media/audio/pulse/pulse_input...
media/audio/pulse/pulse_input.cc:305: // |volume_|.
This needs some mention of the fact that the |normalized_volume| is
intentionally an unused value here.

[no longer working on chromium, 2013-02-26 12:02:51]

Thanks Dale, I am going to land it now.

https://codereview.chromium.org/12310102/diff/1/media/audio/pulse/pulse_input.cc
File media/audio/pulse/pulse_input.cc (right):

https://codereview.chromium.org/12310102/diff/1/media/audio/pulse/pulse_input...
media/audio/pulse/pulse_input.cc:281: stream->volume_ =
static_cast<double>(volume);
On 2013/02/25 18:52:18, DaleCurtis wrote:
> I'd add a comment here that this is thread safe because SetVolume runs under a
> PulseLock.

Done.

https://codereview.chromium.org/12310102/diff/1/media/audio/pulse/pulse_input...
media/audio/pulse/pulse_input.cc:305: // |volume_|.
On 2013/02/25 18:52:18, DaleCurtis wrote:
> This needs some mention of the fact that the |normalized_volume| is
> intentionally an unused value here.

Done.

[no longer working on chromium, 2013-02-26 12:05:51]

Committed patchset #2 manually as r184629 (presubmit successful).